strnad/CrewAI-Studio

CrewAI Studio – a no‑code UI for building and running AI agent crews

What it is – CrewAI Studio is a small web application (built with Streamlit) that lets you create, configure and run crews of AI agents powered by the CrewAI framework. It is aimed at users who want to experiment with multi‑agent workflows without writing Python code.

Key capabilities

• Graphical crew designer – define agents, their tools and the knowledge sources they can use via a point‑and‑click interface.
• Run crews in the background – start a crew, watch its progress, and stop it if needed.
• Result history – previous runs are stored locally and can be inspected later.
• Custom tools – the UI exposes a set of ready‑made tools (API callers, file writers, enhanced code interpreter, web scraper, etc.) and lets you add your own.
• Multiple LLM back‑ends – supports OpenAI, Anthropic, Groq, Ollama, Grok and LM Studio. For most providers you only need the API key; LM Studio also requires you to load an embedding model.
• Cross‑platform – works on Windows, macOS and Linux, and can be installed via Conda, a Python virtual environment, Docker‑Compose, or a one‑click RepoCloud deployment.
• Export – crews can be turned into a single‑page Streamlit app that can be shared or deployed elsewhere.

How to get it running

1. Clone the repo.
2. Choose an installation method:
   • Docker‑Compose – copy .env_example to .env, fill in your API keys, then docker-compose up --build.
   • Conda – run install_conda.sh (or the Windows .bat), then run_conda.sh.
   • Virtual‑env – run install_venv.sh (or the Windows batch), then run_venv.sh.
3. Open a browser at http://localhost:8501 to see the Streamlit UI.

Typical workflow

1. Add knowledge sources (e.g., documents, URLs) that agents can query.
2. Create a crew – add agents, assign tools, set prompts.
3. Select an LLM provider and supply the required API key.
4. Launch the crew; the UI shows live logs and lets you stop the run.
5. Review the output in the history view or export the crew as a standalone app.

Who might use it

• Product managers or researchers who want to prototype multi‑agent pipelines quickly.
• Educators demonstrating agent orchestration concepts without diving into code.
• Developers who prefer a visual front‑end for configuring CrewAI before integrating the generated crew into larger systems.

Project health – The repository is in low‑maintenance mode: no new features are planned, but security and critical bug fixes are still accepted, and pull requests are welcomed.

Support – The author lists a Bitcoin address and a GitHub Sponsors badge for donations.

---

All information above is taken directly from the project's README.